1: \begin{abstract}

2:   The Naming Game is a model of non-equilibriun dynamics for the

3:   self-organized emergence of a linguistic convention or a communication

4:   system in a population of agents with pairwise local interactions. We

5:   present an extensive study of its dynamics on complex networks, that can be

6:   considered as the most natural topological embedding for agents involved in

7:   language games and opinion dynamics. Except for some community structured

8:   networks on which metastable phases can be observed, agents playing the

9:   Naming Game always manage to reach a global consensus.  This convergence is obtained

10:   after a time generically scaling with the population's size $N$ as $t_{conv}

11:   \sim N^{1.4 \pm 0.1}$, i.e.  much faster than for agents embedded on regular

12:   lattices.  Moreover, the memory capacity required by the system scales only

13:   linearly with its size.  Particular attention is given to heterogenous

14:   networks, in which the dynamical activity pattern of a node depends on its

15:   degree. High degree nodes have a fundamental role, but require larger memory

16:   capacity. They govern the dynamics acting as spreaders of

17:   (linguistic) conventions. The effects of other properties, such as the

18:   average degree and the clustering, are also discussed.

19: \end{abstract}

20: